Skip to main content
. Author manuscript; available in PMC: 2017 Apr 1.
Published in final edited form as: J Neurosci Methods. 2016 Feb 5;263:36–47. doi: 10.1016/j.jneumeth.2016.01.022

Table 1.

Comparison of Hamming error (see Eq. 18) computed from the same nine simulated data sets as above. The VB inference methods tend to overestimate the number of states and therefore have much larger Hamming error.

Dataset 1 2 3 4 5
HMM (Gibbs) 9 401 13 24 615
HMM (VB) 166 290 295 123 124
HDP-HMM (Gibbs) 2 3 5 1 6
HDP-HMM (HMC) 3 4 3 2 4
HDP-HMM (EB) 1 3 2 3 12
HDP-HMM (VB) 432 586 340 264 675